Since the early 1980s, some cases of acute and unexplained respiratory distress syndrome which generated several deaths began to be detected in NWA. Later, when HPS and its causal agent were discovered in the USA during the early 90s, it was possible to retrospectively analyze stored blood samples through serological studies and confirm that they were cases of HPS [4]. Despite the almost 30 years since the identification of this syndrome in the area until now, no risk maps were constructed that would allow health authorities to be on a high alert for orthohantavirus infections. In this context, we have sought to develop a HPS risk map from distribution models of known reservoirs for orthohantaviruses using the maximum entropy method to estimate the ecological niche and the potential geographic distribution of these rodent species.
This is relevant to the planning of control and preventive HPS measures, because the knowledge of how climatic and environmental variables determine the distribution and habitat preference of organisms linked to zoonotic diseases can help us identify the geographical risk of disease transmission to the human population based on quantitative criteria [15, 16]. Particularly, here we analyze the three rodent species that have been identified as reservoirs of two orthohantavirus species and three virus strains in NWA [11, 12, 38]. The COVID-19 pandemic has drawn attention to the need for a more in-depth understanding of aspects related to the ecology of rodents that are involved or potentially may become involved in the transmission of zoonotic diseases [39].
As for many organisms, the rodents' abundance can be affected by temperature and precipitation, since this can favor the growth of vegetation and access to food and shelter, thus favoring their reproduction and survival. Also, these ecological relevant climatic factors frequently impose limits to species physiological tolerance and thus their geographical distribution. However, the effect of the climatic variables on all rodent species is not the same [40]. For the three rodents, the variables with the greatest contribution to the models, associated with temperature were temperature seasonality, isothermality (O. chacoensis) and mean annual temperature (O. f. occidentalis and C. fecundus). The average annual temperatures for places with presence of these three rodents varies between 14.7°C to 25.7°C, 6.85°C to 23.2°C and − 1.4°C to 27,7°C respectively. These value ranges in NWA are observed principally in subtropical areas that include the Yunga rainforest and Dry Chaco ecoregions, markedly different from that expected for the O. longicaudatus species, which is also a reservoir of orthohantavirus in southern Argentina [19].
Regarding the rainfalls, our models for the three rodent species showed a greater contribution of the rainfall seasonality variable, detecting that relatively lower seasonal variation in rainfall contributes more to the O. chacoensis model. Besides, an increase in the amount of annual precipitation generates greater suitability for rodent species and if this occurs for orthohantavirus reservoirs, the risk of transmission may be increased. Abrupt increases in population density have been reported in Argentina for the Akodon, Calomys, Mus and Oligoryzomys genera, known as “ratadas”, associated with increases in mean annual rainfall [41]. Similar results were previously reported for reservoirs in the southern endemic region [19, 20] and for C. fecundus in the NWA endemic region [14].
The contribution of the mean NDVI to the distribution models of O. f. occidentalis and C. fecundus may be related to the productivity of the ecosystem in the M area. In fact, a previous study carried out in Mexico, in which rodents’ species were monitored on an altitudinal gradient, highlighted the importance of ecosystem productivity (estimated by NDVI) on the diversity and abundance of rodents [42]. However, vegetation cover can help increase species diversity (including competitors and predators) and thus can also maintain reservoir populations at relatively low abundances which may affect the dynamics of the risk of virus transmission [43].
The steep elevation gradient of the Andes mountains appeared as a natural barrier limiting the western distribution of the three species [14, 44]. However, in the eastern half of the study region, we found differences between the potential geographic distributions predicted by the models for the three reservoirs. Although the three species have high environmental suitability in the Yungas ecoregion, O. chacoensis shows greater environmental suitability followed by C. boliviae, then O. f. occidentalis in the dry Chaco [14]. In addition, the niche overlap was quite low between the species, being O. f. occidentalis the rodent with the largest minimum volume ellipsoid (Fig. 3), which indicates that this species has great adaptive plasticity [22, 45].
The differences found in the suitable areas for these three reservoirs could have important implications for the occurrence of HPS in the northwestern Argentina endemic region. According to our results, the two species involved in the transmission of pathogenic orthohantavirus, O. chacoensis and O. f. occidentalis, for ORNV and BERV respectively, would have similar distribution in the Yungas Forest, but in the Dry Chaco, O. chacoensis would have a predominant role. Probably the combined abundance of the two rodent species can generate a greater number of HPS cases, compared to sites where only one of the species is present. On the other hand, C. fecundus, carrying LNV has suitable habitats in both ecoregions, but only a few infected rodents and human cases of HPS associated with LNV have been recorded in NWA [7].
Our risk map suggests that the areas with the highest risk of orthohantavirus transmission were mainly the Yungas Forest and the occidental zone of Dry Chaco. Particularly, our results highlight the north of Salta province and the east of Jujuy province, where indeed the HPS cases are highly clustered [6, 8]. This suggests that our risk map has a very good sensibility in detecting very high-risk areas for orthohantavirus transmission based solely on the reservoirs distribution modeling. It is important to highlight that most localities with HPS cases are geographically included in the highest stratum (very high-risk) of the risk map and that prevalences/department are positively correlated with risk levels (Fig. 5A y B) [8, 46].
The risk map also predicts areas with high risk where no cases of HPS were reported. However, HPS have been recently reported in the southern area of the NWA within the province of Tucuman [36, 37]. This shows that the risk zones predicted by the risk map have the conditions for the maintenance of orthohantavirus zoonotic transmission cycle, and that the illness is probably under-reported. These places had not previously reported cases and are located in the very high-risk and high-risk strata according to the risk map, a zone that has precipitation conditions similar to the northern sector, where most of the cases are concentrated (considering the latitudinal variation; Fig. 6A and B).
We found that localities with the highest number of cases are located in the northern part of NWA, and that their cases seem to be associated with high rainfall and low seasonality of rainfall (Fig. 6A and B). In a systematic review, evidence was found for the association of HPS with precipitation and habitat type, and mixed evidence for temperature and humidity [40]. Furthermore, Ferro et al., (2020) showed a relationship between HPS cases and lagged rainfall and temperature with a delay of 2 to 6 months in the Argentinean northwestern region.
Assuming there are no geographic barriers, bioclimatic conditions can affect the presence of the virus in a given place in two main ways: a) indirectly, favoring or limiting the distribution and abundance of reservoirs, and b) directly, by affecting the viability of the virus in the environment. The comparison of the risk areas predicted by the risk map (constructed from the potential distribution of reservoirs), with the areas where the sustained presence of HPS cases occur, might indicate which are the limiting bioclimatic variables (and the thresholds) of orthohantavirus distribution.
The high-risk areas of orthohantavirus transmission in the NWA have a subtropical climate with high temperatures and summer rainfall, which favors the development of vegetation that serves as food and shelter for rodents. This in turn gives rise to a population growth of the reservoirs and with this there is a greater probability of contact between them either in burrows or through fights between adult male rodents, which leads to a higher prevalence of viral infection in these reservoirs. In this context, orthohantavirus transmission commonly occurs in wild or rural places when people go to those sites to work, or to carry out recreational activities or tourism. Recreational activities normally include hunting and fishing. Meanwhile, occupational exposure can include deforestation, agriculture, transport and military activities [6, 38]. HPS cases were even detected in ferrymen (or bagayeros) that carry out cross-border trade of legal and illegal merchandise.
Added to this, in rural places transmission events occur when rodents invade the sheds or warehouses indoors and their excretions are aspirated by people when cleaning the place [8]. The generation of dust with viral particles can be favored by lower humidity and therefore by a lower precipitation amount, which can occur in the subtropical zone between autumn and spring months [40]. In addition, the lower temperatures in that period may favor greater preservation of the virus in the environment. Finally, we have found rodents of the genera Oligoryzomys and Calomys on the outskirts of cities in the department of Orán in the study area, which may represent a high risk of transmission in densely populated places (Non published data).
Deforestation normally causes a diversity loss and some reservoir rodents can continue to find shelters on those altered sites or even migrate to populated rural settlements or urban places. It has been previously reported that the loss of diversity in ecosystems can increase the risk of Hantavirus transmission due to the loss of the dilution effect [47]. This dilution effect consists in ecosystems with high diversity where the reservoir-virus-reservoir contact is hindered by their interaction with other organisms [43, 48, 49]. Particularly, in NWA most sites with HPS cases are at a short distance from deforestation areas.
In addition, when we talk about the population at risk, we are considering the areas with high risk according to the model and the departments with a frequent presence of cases. In our opinion, not the entire area of an apartment is at risk of transmission. However, due to a matter of proximity, we can think of risk as the possibility that people may be exposed to risk due to the displacement of people (whether for work, recreation, etc.). The low global prevalence makes us think that the areas with optimal environmental conditions for the virus are not usually very accessible to the global population.
As limitations of this work, we can mention the difficulty to collect complete information of some cases in the region, including those reported at the beginning of the study period. Additionally, it is important to inquire about unknown hosts and reservoirs in the region, since there is increasing evidence of several animals linked to Hantavirus, such as bats or new rodent species, which may act as host and potential reservoirs of these and currently unknown viruses [50]. For example, C. callosus and C. laucha are reservoirs of LNV in Bolivia and Paraguay respectively. Therefore, our risk map should be updated periodically depending on whether new records and reservoirs for this zoonotic disease are found. We must also consider that the possible low accessibility of people to areas of high environmental suitability for the virus may be generating an overestimation of the population at risk.